AI Product Engineer

AI Product Engineer Part time to start, 20 to 30 hours a week. Remote. At least 4 hours a day overlapping with US Eastern business hours. ABOUT PR3CORIA PR3CORIA is an Agentic AI operating system that turns risk intelligence, culture maturity, and high-stakes communication into one integrated operating layer, helping leaders and teams identify signals, strengthen organizational readiness, and act before the moment escalates. It is built on three core intelligences, PR3-RISK, PR3-CULTURE and PR3-COMMS, and we're preparing to welcome our first beta clients. We're a small team working remotely across several time zones. ABOUT THE ROLE We're looking for an AI Product Engineer to join our product team in-house. Picture a business analyst, a software architect and a software developer with deep AI skills, in one person. You'll take technical ownership of our existing platform codebase, work directly with our team to build and refine the product, and turn feedback from our beta clients into features that work. Much of the work happens live: you and our CEO in a working session together, testing a feature, finding what isn't right, fixing it and testing it again. Between sessions, you work independently through the product backlog our Product Lead manages. Organizations will trust PR3CORIA with sensitive information about their risk, culture and communication, so security and quality are built into every decision you make, from the infrastructure to the interface. WHAT YOU'LL DO - Take technical ownership of the existing codebase: get it running in our own cloud environment, understand its dependencies and deployment process, address technical debt, and document it well enough for another engineer to maintain it. - Build with AI coding agents as your everyday way of working, direct them well, check everything they write, and test it before it ships. - Build, fix and ship features in short loops, often live with our CEO, and turn beta client feedback into working features. - Design and maintain our multi-agent AI architecture, including agent orchestration, retrieval (RAG), tenant-aware access controls, traceability and audit logging. - Write, test and refine the prompts and agent instructions behind the product, and build repeatable evaluations to assess accuracy, reliability, regressions and failure cases. Keep AI-generated observations separate from approved decision and scoring rules. - Build security in from the start, working with our cybersecurity lead on authentication, tenant isolation, least-privilege access, data protection, secrets management and secure infrastructure. - Keep quality high: automated unit and integration tests, peer review before production releases, reproducible deployments, and code clean enough for someone else to pick up. - Bridge business and code: clarify requirements and assumptions early, turn them into testable specifications, estimate realistically, flag blockers before they become delays, and propose alternatives when a request creates technical or security risk. - Build front ends that match our designs, because how PR3CORIA looks and feels matters as much as how it functions. - Send a short written update each week covering what shipped, what's next and anything that's blocked. WHAT YOU BRING We're hiring for judgement more than for one language. Our back end is Python, so you need to read it well enough to review it. We don't ask for years in any single language. Must have: - You build software with AI coding agents every day, ideally Claude Code, and can show us how you work. - You review AI-written code critically and catch what is wrong, insecure or untested before it ships, and you'll say no to a request that isn't safe. - AI and LLM applications you've built that are running in production, not only prototypes or demos. We'll ask to see them. If client confidentiality stops you showing the system itself, a sanitized architecture walkthrough with verifiable evidence works. - Hands-on experience with retrieval-augmented generation (RAG), agentic systems, retrieval permissions, and systematic prompt and output evaluation. - Full-stack skills across back end, APIs, databases and front end, with the ability to troubleshoot workflows across the whole stack. - Experience deploying and running applications on Microsoft Azure, including CI/CD and debugging production issues. - A security-first approach: authentication, tenant-level authorization, encryption, least-privilege access, auditability and secrets management are part of how you build. - The ability to take over an existing codebase and maintain it independently, with documentation and testing discipline: unit, integration and regression tests, plus a clear way of verifying production fixes. - A clear understanding of the difference between AI-generated recommendations and deterministic business logic. - The ability to explain technical decisions clearly to non-technical founders, and to understand the business reason behind what you're building. - Excellent written and spoken English. - At least 4 hours a day of overlap with US Eastern business hours. Nice to have: - Experience building with the Anthropic Claude API. - Experience building software in governance, risk, compliance, security or HR, or for regulated industries. - A background in business analysis or product. - Experience working toward a recognized security framework such as SOC 2. WHO YOU ARE At PR3CORIA, trust is our biggest currency. We trust you with your time, your judgment and your voice, and you earn that trust by delivering on what you take on and by telling the truth, especially when it's hard. We're looking for someone who is: - Self-led and self-accountable. You own your responsibilities and your time, and what you take on gets done. - Proactive, not silent. When a blocker or a delay shows up, you say so before anyone has to ask. - Truthful, direct and kind, including when the news is unwelcome or inconvenient. - Growth-minded. You take feedback professionally, without defensiveness, and use it to improve. - A "yes, and" builder. You build on other people's ideas before you critique them, and when something would compromise our standards, you say no openly, explain why and offer what could work instead. - Committed to world-class quality, in the work itself and in how you serve the people who use it. - Interested in what we're building: helping organizations bring risk, culture and communication together. THE ENGAGEMENT - Contract: hourly contract through Upwork. As scopes become clear, parts of the work may move to fixed-price milestones. - Hours: 20 to 30 hours a week to start, tracked through Upwork. - Rate: $60 to $90 an hour, depending on experience. - Location: remote, with at least 4 hours a day of overlap with US Eastern business hours. - Start: as soon as the right person has completed our hiring process. HOW WE HIRE 1. Apply through Upwork, including answers to our screening questions. 2. A screening call with a member of our technical team. 3. A practical skills exercise of about two hours, using a small simulated application and focused on debugging, secure engineering, technical judgment and handover. Use AI coding tools as you normally would, and send us your prompts and the agent's transcript with your answers. 4. An interview with our CEO. 5. Reference checks, with your permission. 6. A paid, fixed-price trial project of roughly 8 to 12 hours, delivering a bounded improvement to our platform with agreed acceptance criteria, tests and handover notes. You'll sign a non-disclosure agreement before you see the code. 7. If the trial goes well, we talk about onboarding and an ongoing contract.